Improved Probabilistic Collaborative Representation-Based Pattern Classification

Pattern recognition,also known as pattern classification,is an important part of information science and artificial intelligence.With the development of science and technology,it has made rapid progress in the past decades.The representation-based classification has attracted much attention in pattern recognition and machine learning.Asatypicaltypeofrepresentation-basedclassification,collaborative representation-based classification is very simple,efficient and robust for classification because of its closed-form solution.In order to further improve the recognition accuracy and classification performance of collaborative representation,we mainly focus on three aspects:With the purpose of further exploring the classification performance of collaborative representation-based classification?CRC?and probabilistic collaborative representation-based classification?PCRC?and making full use of local information to enhance the discrimination of representation coefficients,we propose the residual-based extensions and the weighted ones of CRC and PCRC in this chapter.The proposed residual-based extensions of CRC and PCRC are mainly constrained by jointing thel₁-norm and thel₂-norm of coding residuals on the representation fidelity.In the weighted extensions of CRC and PCRC with different coding residuals,the representation coefficients are constrained by the locality of data.The experimental results have been proved that the proposed extensions of CRC and PCRC are very effective and robust for classification.Inspired by PCRC and the idea of coarse to fine representation,we propose two-phase probabilistic collaborative representation based-classification?TPCRC?to improving the classification performance of PCRC.In TPCRC,the first phase is to utilize probabilistic collaborative representation to coarsely choose the nearest representative samples,and the second phase is to use the chosen nearest samples to finely represent and classify each testing sample.In order to employ the locality of data to improve classification performance PCRC,we propose two-phase weighted probabilistic collaborative representation based-classification?TWPCRC?.A testing sample in most of the CRC variants is collaboratively reconstructed by a linear combination of the training samples from all the classes,the training samples from the class that the testing sample belongs to have no advantage in discriminatively and competitively representing and classifying the testing sample.Moreover,the incorrect classification can easily come into being when the training samples from the different classes are very similar.To address the issues,we propose a novel discriminative collaborative representation-based classification?DCRC?method vial₂regularizations to enhance the power of pattern discrimination.In the proposed model,we consider not only the discriminative decorrelations among all the classes,but also the similarities between the reconstructed representation of all the classes and the class-specific reconstructed representations in thel₂ regularizations.